Overdoses from opioid pain relievers (OPR) have reached epidemic proportions in the United States. In response to this public health crisis, there have been local, state and national efforts to reduce inappropriate opioid prescribing. Recent small reductions in OPR deaths rates in 2012-3 suggest that these efforts may have been effective. However, unprecedented recent increases in heroin use, heroin use disorders, and heroin overdoses have also been observed. It has been suggested that efforts to curb inappropriate OPR prescribing are having the unintended consequence of increasing heroin use and its adverse events. Low cost and wide availability are common reasons provided for the switch from OPR to heroin use. To effectively manage the risk of heroin initiation, health systems, insurance plans, and medical providers need to understand whether, how, and for whom these risks are affected by OPR prescribing practices. Several policies in Colorado have been directed at curbing opioid prescribing. In August of 2014, Colorado Medicaid limited monthly quantities of short-acting opioid pills to 120. In February 2016, Medicaid limited the average daily covered morphine equivalent dose (MED) to 300 MED. Kaiser Permanente Colorado (KPCO) follows Medicaid's lead on opioid policy. These policies were intended to encourage providers to taper patients from high-dose opioid therapy, thus reducing the risk of OPR overdose. To date, however, the effect of these policies on physician prescribing behavior, OPR overdose, heroin use, and heroin overdose is not known. The overall goal of this project is to conduct a large, longitudinal cohort study to examine the impact of OPR policies and prescribing practices on heroin use and heroin overdose from 2012 to 2017. The study will include patients from a large managed care organization (Kaiser Permanente Colorado) and Colorado's Medicaid program. Together these systems cover approximately 2 million people in Colorado, representing 40% of the state's population. Exposures and outcomes, and covariates will be measured by linking patient-level data from pharmacy records, diagnoses, toxicology findings, health care utilization, and state vital records. Exposures and outcomes will be validated with medical record review. Analyses will be conducted with linear and nonlinear mixed effects regression models, with propensity scores to account for confounding. The results of these analyses will have implications for OPR prescribing policies at the state and national level.